JM - I'm with Albert on this - it's some corruption in the data within a specific table. R:Scope would not necessarily find that problem.
If Albert's suggestion still doesn't show you the offending records, do this: To try to narrow it down, I'd modify the SELECT statement to get just the first half of the expected records - if that works you know the problem is in the second half of the table, if it bombs you know the problem is in the first half of the table. Then I'd continue restricting the SELECT statement until I found the set of records that cause the abort. Next, unload the data from that table to a file and open the file in a text editor and search for records around the offending record. If you scroll down and up around that area you will hopefully see high ASCII heiroglyphics and will be able to see which record needs to be deleted from the database. Good luck! Sami ----------------------------------------------------------------- Sami Aaron Software Management Specialists 19312 W 63rd Terr Shawnee KS 66218 mailto:[EMAIL PROTECTED] 913-915-1971 ----- Original Message ----- From: "J.M. GRATIAS" <[EMAIL PROTECTED]> To: "RBASE-L Mailing List" <[EMAIL PROTECTED]> Sent: Saturday, November 08, 2003 4:07 PM Subject: [RBASE-L] - RE: RB 4.5++ abort messages Albert : > JM, I think it is time to R:Scope the database. One of the records covered by the select has high order ascii characters in it, and when R:Base tries to read the record, it blows up. An autochk will not necessarily find the error if it is data and does not alter the row length. < I already did check the DB : RScope/Autochk did't report any troubles into the DB structure. What I suspect is a brooken index. Is there a way to trap or see RBase abort message before return to OS ? J.M. GRATIAS, Logimatique
